Identification of new biomarkers for human papillary thyroid carcinoma employing NanoString analysis

Abstract

We previously reported an upregulation of the clock transcript BMAL1, correlating with TIMP1 expression in fresh-frozen samples from papillary thyroid carcinoma (PTC). Since frozen postoperative biopsy samples are difficult to obtain, we aimed to validate the application of high-precision NanoString analysis for formalin-fixed paraffin-embedded (FFPE) thyroid nodule samples and to screen for potential biomarkers associated with PTC. No significant differences were detected between fresh-frozen and FFPE samples. NanoString analysis of 51 transcripts in 17 PTC and 17 benign nodule samples obtained from different donors and in 24 pairs of benign and PTC nodules, obtained from the same donor (multinodular goiters), confirmed significant alterations in the levels of BMAL1, c-MET, c-KIT, TIMP1, and other transcripts. Moreover, we identified for the first time alterations in CHEK1 and BCL2 levels in PTC. A predictive score was established for each sample, based on the combined expression levels of BMAL1, CHEK1, c-MET, c-KIT and TIMP1. In combination with BRAF mutation analysis, this predictive score closely correlated with the clinicopathological characteristics of the analyzed thyroid nodules. Our study identified new thyroid transcripts with altered levels in PTC using the NanoString approach. A predictive score correlation coefficient might contribute to improve the preoperative diagnosis of thyroid nodules.

Keywords: FFPE; NanoString analysis; biomarkers; circadian clock; papillary thyroid carcinoma.

Figure 1. Pair-wise correlations between BMAL1, CHEK1, TIMP1, c-KIT and c-MET transcript changes in PTC

Pearson correlation analysis revealed pair-wise correlations of gene expression in PTC (combined samples from 1^st and 2^nd cohorts, Tables 1 and 2) and benign (1^st cohort, Table 1) samples. The correlation strength was based on Evans classification (see Methods for details), with a coefficient r value < 0.20 reflecting very weak correlation; 0.20 – 0.39 – weak; 0.40 – 0.59 – moderate; 0.60 – 0.79 – strong; and > 0.80 - very strong. The dots at each graph are representing normalized respective gene expression values.

Figure 2. Scatter plot of gene expression-based predictive scores correlated to PTC biological aggressiveness and BRAF mutation status

The gene expression-based score for benign and PTC samples was calculated based on joint expression levels of BMAL1, CHEK1, c-KIT, c-MET and TIMP1 transcripts. Thus obtained score values were plotted for three clusters: benign, less aggressive PTC and more aggressive PTC, and allowed a clear distinction between these groups. BRAF^V600E mutation was strongly associated with samples characterized by the highest predictive score values and with more aggressive PTC diagnosis.